Google Unveils Lyria 3 Music Generator for 30-Second AI Tracks

Google has officially launched Lyria 3, its latest generative music model, as a beta feature within the Gemini app, allowing users to create 30-second audio tracks from text prompts or visual inputs. The announcement, detailed in an official Google blog post, positions the tool as a creative expression aid rather than a replacement for professional music production.

Free Gemini users will be limited to 30-second tracks—comparable to iTunes preview lengths—while paid subscribers (AI Plus, Pro, and Ultra tiers) receive extended durations, though Google has not specified exact limits. The model supports eight languages at launch: English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese, with additional languages planned for future updates. Users must be 18 or older to access the feature, per the company's terms.

Lyria 3 represents a significant evolution from previous iterations, which were restricted to Google's Music AI Sandbox for professional musicians. The new model automates lyric generation—a key improvement over earlier versions that required user-provided text—and offers granular control over musical elements like tempo, vocal style, and instrumentation. Google claims the model produces "more realistic and musically complex tracks," though the company acknowledges outputs are designed for casual use rather than artistic mastery.

Users can generate tracks through text prompts (e.g., "a comical R&B slow jam about a sock finding their match") or by uploading images/videos for visual-inspired compositions. The system also creates AI-generated album art via Google's Nano Banana model, with all outputs embedded with SynthID—a watermarking technology Google developed to identify AI-generated content. This aligns with the company's broader commitment to responsible AI deployment, as noted in the blog post.

Per Google's documentation, Lyria 3 operates through the Gemini API with two distinct model variants: Lyria 3 Clip (fixed 30-second outputs) and Lyria 3 Pro (for longer tracks with customizable structure). The API documentation confirms the model produces 44.1 kHz stereo audio with structural coherence, including timed lyrics and full instrumental arrangements. However, the company has not yet integrated Lyria 3 Pro into the public Gemini app, focusing first on the clip-based feature.

The feature's limitations reflect Google's cautious approach to AI music generation. As Gizmodo reported, the 30-second cap prevents users from creating full-length songs, and the company explicitly states the goal is "to give you a fun, unique way to express yourself" rather than produce "musical masterpieces." This contrasts with competitors like Suno AI, which offers longer tracks but faces greater copyright scrutiny.

Google's implementation also addresses industry concerns about AI's impact on music creators. The company emphasizes that Lyria 3 is "designed for original expression, not for mimicking existing artists," with filters to prevent outputs from closely resembling copyrighted works. Users can report content violating intellectual property rights, and all outputs must comply with Google's Terms of Service and Gen AI prohibited use policies.

For developers, the Lyria 3 API provides a standardized interface for integrating music generation into third-party applications. The API documentation includes code samples for Python, JavaScript, and Java, enabling developers to generate tracks programmatically. However, the current public release remains limited to the Gemini app, with YouTube's "Dream Track" feature set to incorporate Lyria 3 for Shorts soundtrack customization.

Industry analysts note that Google's focus on short-form, shareable tracks aligns with social media trends. As Engadget observed, the 30-second format suits platforms like TikTok and Instagram, where users frequently share brief audio snippets. The integration of album art via Nano Banana further streamlines the sharing experience, eliminating the need for manual design.

While Lyria 3's technical capabilities represent a step forward in AI music generation, its practical utility remains constrained by the 30-second limit. The model's ability to generate coherent lyrics and instrumentals—evident in Google's promotional examples—suggests potential for creative experimentation, though the company's own disclaimer about "corny and strange" lyrics hints at current limitations.

Google's approach reflects a broader industry trend of prioritizing responsible AI deployment over rapid feature expansion. By embedding SynthID and restricting outputs to short durations, the company aims to balance innovation with ethical considerations—a strategy that may influence how other tech giants approach generative music tools in the future.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn